Variation of chemical compounds in wild reveals ecological factors involved in the evolution of chemical defenses in mimetic

Sculfort, Ombeline; de Castro, Erika C. P.; Kozak, Krzysztof M.; Bak, Søren; Elias, Marianne; Nay, Bastien; Llaurens, Violaine

Published in: Ecology and Evolution

DOI: 10.1002/ece3.6044

Publication date: 2020

Document version Publisher's PDF, also known as Version of record

Document license: CC BY

Citation for published version (APA): Sculfort, O., de Castro, E. C. P., Kozak, K. M., Bak, S., Elias, M., Nay, B., & Llaurens, V. (2020). Variation of chemical compounds in wild Heliconiini reveals ecological factors involved in the evolution of chemical defenses in mimetic butterflies. Ecology and Evolution, 10(5), 2677-2694. https://doi.org/10.1002/ece3.6044

Download date: 27. sep.. 2021

Received: 29 November 2019 | Revised: 6 January 2020 | Accepted: 7 January 2020 DOI: 10.1002/ece3.6044

REVIEW

Variation of chemical compounds in wild Heliconiini reveals ecological factors involved in the evolution of chemical defenses in mimetic butterflies

Ombeline Sculfort1,2 | Erika C. P. de Castro3 | Krzysztof M. Kozak4 | Søren Bak5 | Marianne Elias1 | Bastien Nay2,6 | Violaine Llaurens1

1Institut de Systématique, Evolution, Biodiversité (ISYEB), Muséum National Abstract d'Histoire Naturelle, CNRS, Sorbonne- Evolutionary convergence of color pattern in mimetic species is tightly linked with Université, EPHE, Université des Antilles, Paris, France the evolution of chemical defenses. Yet, the evolutionary forces involved in natural 2Unité Molécules de Communication et variations of chemical defenses in aposematic species are still understudied. Herein, Adaptations des Micro-organismes (MCAM), we focus on the evolution of chemical defenses in the tribe Heliconiini. Muséum National d'Histoire Naturelle, CNRS, Paris, France These neotropical butterflies contain large concentrations of cyanogenic glucosides, 3Department of Zoology, Cambridge cyanide-releasing compounds acting as predator deterrent. These compounds are ei- University, Cambridge, UK ther de novo synthesized or sequestered from their Passiflora host plant, so that their 4Smithsonian Tropical Research Institute, Panamá, Panamá concentrations may depend on host plant specialization and host plant availability. 5Department of Plant and Environmental We sampled 375 wild Heliconiini butterflies across Central and South America, cov- Sciences, University of Copenhagen, ering 43% species of this clade, and quantify individual variations in the different CGs Frederiksberg, Denmark 6Laboratoire de Synthèse Organique, Ecole using liquid chromatography coupled with tandem mass spectrometry. We detected Polytechnique, CNRS, ENSTA, Institut new compounds and important variations in chemical defenses both within and Polytechnique de Paris, Palaiseau Cedex, France among species. Based on the most recent and well-studied phylogeny of Heliconiini, we show that ecological factors such as mimetic interactions and host plant spe- Correspondence Ombeline Sculfort, Institut de Systématique, cialization have a significant association with chemical profiles, but these effects are Evolution, Biodiversité (ISYEB), Muséum largely explained by phylogenetic relationships. Our results therefore suggest that National d'Histoire Naturelle, CNRS, Sorbonne-Université, EPHE, Université des shared ancestries largely contribute to chemical defense variation, pointing out at Antilles, 45 rue Buffon, 75005 Paris, France. the interaction between historical and ecological factors in the evolution of Müllerian Email: [email protected] mimicry. Funding information European Research Council, Grant/Award KEYWORDS Number: 339873 ; Marie Curie Actions; aposematism, cyanogenic glucosides, Heliconius, LC-MS/MS, Müllerian mimicry, phylogenetic Agence Nationale de la Recherche, Grant/ Award Number: ANR-10-LABX-0003-BCDiv signal and ANR-11-IDEX-0004-02

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2020 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd.

 www.ecolevol.org | 2677 Ecology and Evolution. 2020;10:2677–2694. 2678 | SCULFORT et al.

1 | INTRODUCTION chemical defenses in mimetic butterflies can result from complex interactions between host plant adaptation and predation pressure. The evolution of complex phenotypes combining different traits A recent survey of natural populations of two comimetic butterfly subject to natural selection raises the question of the mechanisms species, the viceroy (Limenitis archippus) and queen (Danaus gilippus), underlying adaptation involving multiple traits. In aposematic spe- demonstrated that the average concentration of chemical defenses cies for instance, the defensive traits such as toxicity, and the warn- increases in the viceroy populations where the defended queen spe- ing coloration may evolve asynchronously and can be submitted to cies is absent (Prudic, Timmermann, Papaj, Ritland, & Oliver, 2019). contrasted selective pressures. While the evolution of color patterns This effect is independent from variation in defensive compounds and the selective mechanisms involved have received considerable concentrations in the host plants (Prudic et al., 2019), highlighting attention (Le Poul et al., 2014; Sherratt, 2008), the evolutionary or- that the abundance of comimics may modulate selection exerted on igin of chemical defense variations is still understudied. The effect chemical defenses in mimetic species. of chemical defenses on predator avoidance is critical for prey sur- Here, we aim to disentangle the mechanisms involved in the vival (Ihalainen, Lindström, & Mappes, 2007) and therefore central evolution of chemical defenses, from neutral divergence to selec- in the evolution of warning colorations (Blount, Speed, Ruxton, & tive pressure of predation and host plant adaptation. We focus Stephens, 2009; Speed & Ruxton, 2007). By sampling aposematic on the butterflies belonging to the neotropical tribe Heliconiini prey, predators learn to associate deterrent effect with a given warn- (: ), where color pattern evolution and ing color pattern and subsequently avoid any resembling prey item mimetic interactions have been extensively documented (Joron & (Alcock, 1970a, 1970b; Goodale & Sneddon, 1977). The immediate Iwasa, 2005; Joron & Mallet, 1998; Merrill et al., 2015). Subspecies and long-term effect of defensive compounds thus determines the of Heliconiini are defined based on variation in color pattern be- protection gained from aposematism (Skelhorn & Rowe, 2005), and tween geographic locations, observed within species (Braby, therefore the evolution of color patterns. Eastwood, & Murray, 2012). Heliconiini butterflies contain a wide Evolutionary convergence in aposematic signal among co-occur- diversity of defensive compounds, especially aliphatic or cyclopen- ring defended prey species is frequently observed among sympatric tenoid CGs (CGs) (Figure 1) (Castro, Zagrobelny, et al., 2019; Engler, aposematic species, because sharing a color pattern decreases indi- Spencer, & Gilbert, 2000). CGs are supposed to have a bitter and vidual predation risk (Müller, 1879). This results in so-called mimicry repulsive taste (Nahrstedt & Davis, 1985). Additionally, CGs release rings, composed of multiple species sharing a similar warning color toxic cyanide and chemical by-products for birds when put in con- pattern. Both the defensive compounds and the abundance of indi- tact with specific degrading enzymes (Cardoso, 2019; Conn, 1980). viduals sharing a given warning color pattern determine the preda- CGs and enzymes or stored in different cell or tissue compartment tion risk associated with this coloration (Sherratt, 2008). Substantial and are mixed upon tissue disruption under a predator's attack, so quantitative variation in chemical defenses is observed between mi- that Heliconiini butterflies often survive an attack after being tasted metic species, as demonstrated for instance in poison frogs (Santos (e.g., by lizard (Boyden, 1976) or avian predators (Boyden, 1976; Chai, & Cannatella, 2011), marine gastropods opisthobranchs (Cortesi & 1996; Pinheiro & Campos, 2019)). Therefore, the bitter taste pro- Cheney, 2010), or (Arias et al., 2016; Bezzerides, McGraw, vided by CG and toxic metabolites may act as a chemical defense Parker, & Husseini, 2007; Castro, Zagrobelny, et al., 2019). Less de- because of immediate deterrent effect on predator. fended individuals may act as parasites on better defended individ- Heliconiini caterpillars feed on Passiflora plants (Engler & uals by limiting predator avoidance (Rowland, Mappes, Ruxton, & Gilbert, 2007; Jiggins, 2016; Turner, 1967), with substantial be- Speed, 2010; Speed, 1993). The evolution of chemical defenses in havioral variation between species in female egg-laying prefer- mimetic species is thus likely to be influenced by the local abundance ences and in larval survival on different Passiflora species (Benson, of the mimicry ring they belong too, as well as variations in toxin Brown, & Gilbert, 1975; Brown, 1981). Around 30 different CGs levels across individuals composing the ring. have been identified in Passiflora (Castro, Zagrobelny, et al., 2019; Nevertheless, other local ecological factors may influence the Spencer & Seigler, 1987). Larvae of most Heliconiini species syn- evolution of chemical defenses in mimetic species. In butterflies thesize CGs de novo (Wray, Davis, & Nahrstedt, 1983), but many for instance, deterrent compounds, as well as precursors for their sequester CGs from the host plants (Engler et al., 2000). Both syn- synthesis, can be acquired by caterpillars during feeding on specific thesis and sequestration of CGs are only observed in Zygaenidae host plants (Jones, Petschenka, Flacht, & Agrawal, 2019; Nishida, (burnet moths) and Heliconiini, two clades where aposematic color 2002). Chemical defenses may thus vary among species depending patterns have evolved (Zagrobelny, Castro, Møller, & Bak, 2018). on their diet (Engler & Gilbert, 2007). For instance, monarch but- So far, Heliconiini have been reported to sequester five cyclopen- terflies (Danaus plexippus) sequester cardenolides from milkweeds tenoid CGs from Passiflora; the diastereoisomers tetraphyllin B during the larval stage and are thus unpalatable to birds (Brower, and epivolkenin, tetraphyllin A, gynocardin, and dihydrogynocar- McEvoy, Williamson, & Flannery, 1972). Adaptation to host plants is din (Figure 1) (Castro, Zagrobelny, et al., 2019; Engler et al., 2000). thus a key evolutionary factor in the origin and evolution of chem- Heliconiini butterflies can synthesize aliphatic CGs, linamarin, and ical defenses in aposematic butterflies. Nevertheless, because lotaustralin (Figure 1) from the amino acids valine and isoleucine, of the strength of predation on adult butterflies, the evolution of respectively (Nahrstedt & Davis, 1985). Identifying the different SCULFORT et al. | 2679

FIGURE 1 CGs identified in Heliconiini. Framed molecules are aliphatic CGs synthesized by Heliconiini, followed by cyclopentenoid CGs sequestered from Passiflora plants. Glucose group is symbolized by “Glu.” For the first time in Heliconiini, we report epilotaustralin and a stereoisomer of tetraphyllin A (putatively the deidacline, which is not represented here because it was not firmly identified during this study)

CGs may thus allow tracking down their metabolic origins, al- of individuals facing natural variations in host plant availability, mi- though aliphatic linamarin and lotaustralin can also be uptaken metic community abundance, and predator communities. Using liq- by caterpillars, as recently demonstrated in Heliconius melpomene uid chromatography coupled to mass spectrometry (LC-MS/MS), (Castro, Demirtas, et al., 2019). The balance between sequestra- we investigate both quantitative and qualitative variation across tion from host plants and de novo synthesis of CGs in different individuals and then use comparative methods to disentangle phylo- species may be linked to host plant specialization. CG sequestra- genetic and ecological factors influencing the evolution of chemical tion might be more important than synthesis in specialist species, defenses in Heliconiini. as for instance in the specialist species Heliconius sara and H. sapho containing drastically diminished CG concentrations when reared on Passiflora species other than their specific host plants (Engler 2 | MATERIALS AND METHODS & Gilbert, 2007). Evolution of chemical defenses in the Heliconiini clade can thus be influenced by the adaptation to host plants. 2.1 | Butterfly collection The substantial geographic variation in color patterns and host plants observed in the Heliconiini clade (Jiggins, 2016) provides a We sampled butterflies throughout Heliconiini distribution to col- relevant opportunity to investigate the effect of selection pressure lect the maximum number of species. Wild butterflies were caught on the evolution of chemical defenses in mimetic species. Based on from 2016 to 2018 across Peru (n = 286), Panama (n = 45), Ecuador the well-studied phylogeny of Heliconiini (Kozak et al., 2015), we (n = 24), and Brazil (n = 20), using a hand net. We used 375 individu- thus explored how phylogenetic history, mimetic interactions, and als from 33 species, covering 43% of the Heliconiini tribe (Appendix host plant use can drive the evolution of chemical defense in wild 1), and 55 subspecies (Table 1). Individuals were killed by freezing on butterflies. We sampled butterflies throughout Heliconiini distribu- the day of capture (approximately −18°C). Wings were cut at their tion, from Central to South America, in order (a) to maximize the attachment point to the body and preserved dried in an envelope diversity of species of the Heliconiini clade (we cover almost half of and placed in a silica gel containing box to absorb humidity. In order the tribe diversity), and (b) to assess variation in chemical defenses to preserve the integrity of CG molecules, bodies were conserved 2680 | SCULFORT et al.

TABLE 1 Subspecies are divided in nine mimicry rings TABLE 1 (Continued)

Mimicry ring Subspecies Mimicry ring Subspecies

Blue Heliconius congener congener Rayed yellow Heliconius hewitsoni Heliconius doris doris Heliconius pachinus Heliconius doris viridis (blue morph) Heliconius sara magdalena Heliconius sara sara Heliconius wallacei flavescens Tiger dissoluta Dennis ray Eueides tales calathus Eueides isabella hippolinus Heliconius aoede cupidineus Eueides lampeto acacetes Heliconius burneyi jamesi Heliconius ethilla aerotome Heliconius demeter joroni Heliconius hecale felix Heliconius erato emma Heliconius numata arcuella Heliconius eratosignis ucayalensis Heliconius numata lyrcaeus Heliconius melpomene aglaope Heliconius numata tarapotensis Heliconius timareta timareta Heliconius numata zobryssi Heliconius xanthocles melior Heliconius pardalinus butleri Heliconius xanthocles zamora Heliconius pardalinus sergestus Green Philaethria diatonica Other Heliconius melpomene amaryllis X Philaethria dido dido aglaope Philaethria dido panamensis Eueides isabella eva Heliconius charithonia vazquezae Heliconius doris viridis (red morph) Orange Agraulis vanillae luciana Heliconius eleuchia primularis Agraulis vanillae vanillae Heliconius erato cyrbia Dione juno huascuma Heliconius hecale melicerta Dione juno miraculosa Heliconius hecale zuleika Dryadula phaetusa Heliconius numata bicoloratus

Dryas iulia moderata Note: Geographically isolated, phenotypically unique, and hybrid aliphera individuals were assigned to “Other.” Subspecies belonging to the same mimicry ring share a given colour pattern within the same locality. Eueides lybia lybia Mimicry rings and subspecies within are listed in alphabetical order. Postman Panama Heliconius erato demophoon Heliconius melpomene rosina in a plastic vial containing 100% methanol and kept in freezer (ap- proximately −18°C).

Postman Ecuador/Peru Heliconius erato favorinus 2.2 | Cyanogenic glucoside extraction in methanol Heliconius melpomene amaryllis X aglaope For each butterfly specimen, the butterfly body and the methanol Heliconius telesiphe sotericus medium were transferred in a glass tube. Methanol was evapo- Heliconius timareta thelxinoe rated at room temperature until the tissue was fully dried using Savant Automatic Environmental SpeedVac System AES1010 with Postman reverse Heliconius himera VaporNet. For each specimen, body and wings were weighed be- Heliconius timareta timareta fore being crushed together into a fine powder in a glass mortar and pestle using liquid nitrogen. Two mL of 100% methanol was added to the powder before stirring for 1 hr at room tempera- ture. Extracts were centrifugated for 20 min at 4,500 g, filtered using 7 mm diameter glass pipettes and cotton, filtered again with (Continues) a MultiScreen 0.45 µm hydrophilic, low protein binding plate, and SCULFORT et al. | 2681 centrifuged five minutes at 10,000 g. Raw filtrates were diluted chromatogram (EIC) peak areas, and on a regression calculated 50 times in milliQ water, vortexed, and stored in fridge until liq- from the standard curve (in ng of CG/mL of butterfly extract). We uid chromatography and tandem mass spectrometry (LC-MS/MS) reported the concentration of each CG in every butterfly in µg of injections. CG/mg of dried butterfly weight.

2.3 | Liquid chromatography and tandem mass 2.5 | Statistical and comparative analyses spectrometry For each individual, we obtained the concentration of each of the The protocol used in this study has been previously optimized to nine studied CGs, referred to as the chemical profile. By adding identify and quantify CGs in butterfly methanol filtrates (Briolat, these nine CG concentrations, we computed the total CG concen- Zagrobelny, Olsen, Blount, & Stevens, 2019; Castro, Zagrobelny, tration per individual, as an estimation of the amount of chemical et al., 2019). Analytical LC-MS/MS was performed using an Agilent defenses per individual. All statistics were conducted in R 3.4.4 (R: 1,100 Series LC (Agilent Technologies, Germany) coupled to a High The R Project for Statistical Computing, 2019a) and RStudio 1.1.463 Capacity Trap-Ultra ion trap mass spectrometer (Bruker Daltonics, (RStudio, 2019b). Plots were created with ggplot2 3.0.0 package Germany). Chromatographic separation was carried out on a Zorbax (Wickham et al., 2019). SB-C18 column (Agilent; 1.8 μM, 2.1 × 50 mm). Mobile phase A was composed by deionized water containing 0.1% (v/v) formic acid. Mobile phase B was acetonitrile supplemented with 50 μM NaCl 2.5.1 | Qualitative and quantitative variation in CGs and 0.1% (v/v) formic acid. The gradient was as follows: 0–0.5 min, isocratic 2% B; 0.5–7.5 min, linear gradient 2%–40% B; 7.5–8.5 min, We used MANOVA (Multivariate ANalysis Of Variance) to test linear gradient 40%–90% B; 8.5–11.5 isocratic 90% B; 11.6–17 min, whether the (multivariate) CG profiles were different between isocratic 2% B. Flow rate was set to 0.2 ml/min and increased to groups (genera, species, and subspecies), and we reported the name 0.3 ml/min between 11.2 and 13.5 min. During the liquid chroma- of the test, Pillai's trace, degree of freedom, and associated p-value. tography step, initially neutral CGs were associated with Na+ cations We used the Pillai's test because of its robustness regarding hetero- and analyzed with mass spectrometer in the positive electrospray geneities in variance–covariance. mode. The oven temperature was fixed at 35°C. We used ANOVA (ANalysis Of Variance) to test whether the In addition to the 375 butterfly samples, we ran blank control concentration of a specific CG was different between groups. We sample and a reference sample. Blank was methanol gone through presented statistical result of ANOVA as follow: name of the test, the whole protocol extraction, and the reference sample was a mix F value (variance of the group means/ mean of the within group of every butterfly filtrates. CGs were identified by comparison to variances), degree of freedom, and associated p-value. In case of standard solutions (aliphatic were chemically synthesized at PLEN, a significant ANOVA (p-value < .050), post hoc test Tukey Honest Møller, Olsen, & Motawia, 2016, cyclopentenoid were donated by Significant Differences (Tukey's HSD) was done to determine Lawrence Gilbert and Helene Engler, Engler et al., 2000). We made which group was significantly different from the others. Statistical three calibration curves based on three commercial standards: tests were run with R package stats 3.4.2. Heatmap of CG occur- linamarin, lotaustralin/epilotaustralin, and amygdalin (commercial, rence and concentration was plotted using R packages ape 5.1 and Sigma Aldrich), from 0.1 to 20 ng/µl each. Blanks, standards, cali- ggtree 1.10.5 (Paradis, 2011; Yu, Smith, Zhu, Guan, & Lam, 2017). bration curve, and reference sample were run first. The reference sample was injected every ten butterfly samples. 2.5.2 | Evolution of cyanogenic glucoside profiles in Heliconiini 2.4 | Chemical data analyses We calculated the phylogenetic signal of CG profile, that is, the ex- Mass spectra were analyzed using the software Bruker Compass tent to which trait values are explained by the phylogeny, or how DataAnalysis 4.3 (x64). We targeted sodium adducts [M + Na+] much closely related species resemble one another in terms of CG of linamarin [retention time (RT) 2.4 min at m/z 270], lotaustralin profile (Blomberg, Garland, & Ives, 2003). We computed the Kmult [RT 5.4 min at m/z 284], epilotaustralin [RT 5.5 min at m/z 284], statistic, a multivariate extension of Blomberg's K test for univari- tetraphyllin B [RT 1.3 min at m/z 310], epivolkenin [RT 2.3 min at ate phylogenetic signal (Adams, 2014; Blomberg et al., 2003). A low m/z 310], tetraphyllin A [RT 4.9 min at m/z 294], gynocardin [RT phylogenetic signal (Kmult close to 0) indicates a low influence of the 1.4 min at m/z 326], dihydrogynocardin [RT 1.4 min at m/z 328], phylogenetic relationships on the tested trait, whereas high value and amygdalin [RT 6.4 min at m/z 480] (Briolat et al., 2019; Castro, (Kmult close to 1) suggests that the trait evolution along the phylogeny Zagrobelny, et al., 2019). For every targeted CG compound, the is close to Brownian motion. The multivariate phylogenetic signal of total concentration was estimated based on the extracted ion quantitative CG variation across species was evaluated using Kmult 2682 | SCULFORT et al. in the geomorph 3.0.7 R package. We calculated the phylogenetic We investigated variation in total CG concentration, putatively signal in the whole Heliconiini tribe, in the largest genus of the radia- synthesized CG concentration, and putatively sequestered CG tion: Heliconius and more specifically in ancient nodes (pupal-mating concentration between generalist and specialist subspecies. When and nonpupal-mating clades). In Heliconius, phenotypic races of the considering the entire range of a given species across Central and same species often belong to different mimicry rings. Therefore, we South America, it turns out it can have a lot of host plant spe- estimated the phylogenetic signal using mean CG concentrations cies. For instance, Agraulis vanilla has 50 reported host plants and separately at the taxonomic level of species (n = 33) and subspecies Heliconius numata 30 (Kozak, 2016). We conducted our analysis at (n = 55). We adapted the Heliconiini phylogenetic tree (Kozak et al., the subspecies level because locally subspecies actually use much 2015) by pruning species not represented in our sample set. In many less host plants. In our study, generalist is subspecies that feed cases, several subspecies were sampled (for example: H. hecale felix, on more than 5 host plant species whereas specialist subspecies H. hecale melicerta, and H. hecale zuleika). For the subspecies-level feed on 5 or less host plant species. We adjusted this classification analysis, we extended the original phylogeny to include relevant based on the literature. subspecies as follows: the terminal branch length was set equal to the decimal of the previous branch, and the common branch equal to the integer part. All subspecies had same total branch length. In 3 | RESULTS the case of more than two subspecies, the topology was arbitrary resolved. 3.1 | Large variations in the concentration of neo- synthesized and sequestered CGs in wild Heliconiini

2.5.3 | Phylochemospace Across the 375 analyzed Heliconiini samples, nine CGs were identi- fied and important variation in the CG profile was detected between We applied the concept of phylomorphospace, describing mor- genera and species (Table 2). Important variation of CG profile was phological variation across species in correlation with their also detected within species, notably among different subspecies 49 phylogenetic relationships (Sidlauskas, 2008). We built a “phy- (MANOVA, Pillai303 = 3.513, p < .001). lochemospace” describing variation in concentration of multiple Regarding putatively synthesized aliphatic CGs, linamarin was compounds with a principal component analysis (PCA), superim- detected in all 32 out of 33 species, whereas lotaustralin was in posing the phylogenetic relationships among subspecies. The re- all species (Figure 2). However, the concentration of linamarin F32 sulting PCA visualizes the variation in CGs actually occurring in the was significantly different between species (ANOVA, 342 = 13.77, 55 subspecies. Packages FactoMineR 1.41 (Lê, Josse, & Husson, p < .001), and individuals from the genus Eueides had statistically 2008), missMDA 1.14 (Josse & Husson, 2016), and phytools 0.6-44 significant higher linamarin concentration compared with other F6 (Revell, 2012) were used. genera (ANOVA, 368 = 35.46, p < .001; Tukey's HSD, p < .001). Similarly, lotaustralin concentrations differed among species F32 (ANOVA, 342 = 4.324, p < .001). Another aliphatic CG, epilotaus- 2.5.4 | Variation among comimetic subspecies and tralin, was detected in Heliconius, Eueides, Dione, Agraulis, and host plant specialization Dryas genera, with significant variation in concentration among F32 species (ANOVA, 342 = 2.618, p < .001). These three putatively We tested for differences between groups: mimicry ring, geographical synthesized CGs were found at the highest levels in H. charithonia, range, and host plant specialization. We used MANOVA and ANOVA which also did not contain any putatively sequestered CGs in the to assess differences in CG profile and specific CG concentrations, re- two analyzed individuals. spectively, both at species (n = 33) and subspecies (n = 55) level. We Six putatively sequestered CGs from Passiflora host plants were applied Bonferroni correction as we performed several tests on the measured: tetraphyllin A, a diastereoisomer of tetraphyllin A, tet- same dataset. We used stats 3.4.2 for MANOVA and RVAideMemoire raphyllin B, a diastereoisomer of tetraphyllin B called epivolkenin, 0.9-72 package (Hervé, 2019) for associated post hoc test. ANOVA, gynocardin, and dihydrogynocardin. The diastereoisomer of the tet- associated post hoc test, and Bonferroni correction were computed raphyllin A could be deidaclin, because this molecule is also produced with stats 3.4.2 package as well. by Passiflora species used as host plant by Heliconiini butterflies To assess whether the observed statistically significant differ- (Jaroszewski et al., 2002; Spencer, Seigler, & Domingo, 1983; Tober ences were due to shared ancestry, we computed phylogenetic & Conn, 1985). We also searched for the aromatic CGs amygdalin MANOVA and ANOVA, using geiger 2.0.6 (Harmon, Weir, Brock, as it has been measured in few analyzed Passiflora species (Castro, Glor, & Challenger, 2008) and phytools 0.6-44 packages (Revell, Zagrobelny, et al., 2019; Chassagne, Crouzet, Bayonove, & Baumes, 2012), respectively. Phylogenetic MANOVA was performed using 1996), but we did not find aromatic CGs in Heliconiini butterflies, the modified tree and mean CG concentrations per subspecies (as as previously reported in reared H. melpomene (Castro, Demirtas, these phylogenetical tests do not handle multiple values for one sub- et al., 2019). The diversity of putatively sequestered CGs and species, we used mean concentrations). their important variations between species in the wild (MANOVA, SCULFORT et al. | 2683 (Continues) 0.23 0.21 ± ± 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.03 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.04 0.00 ± 0.390.17 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 Dihydrogynocardin 0.00 0.00 0.00 0.00 0.00 0.00 ± 0.61 0.13 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 Gynocardin 4.34 ± 1.86 0.00 0.00 0.00 0.00 Tetraphyllin A stereoisomer 0.77 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.27 ± 1.67 1.89 ± 2.61 0.22 ± 0.50 0.00 0.00 0.31 ± 1.37 0.00 0.00 0.58 ± 2.35 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.05 0.00 0.00 0.00 0.00 0.00 0.48 ± 0.68 0.00 0.00 0.00 0.00 0.00 ± 2.042.11 14.30 ± 26.8515.74 12.34 ± 6.92 Tetraphyllin A 0.07 10.02 ± ± ± 10.567.50 5.65 ± 8.53 0.00 0.00 0.00 0.03 0.00 0.00 0.00 0.00 0.00 0.00 0.93 ± 4.04 0.00 0.00 0.00 0.00 0.73 ± 3.60 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 6.07 31.04 ± 14.70 13.49 ± 18.06 ± 22.4825.96 30.54 ± 8.74 Epivolkenin 30.45 ± 3.79 0.40 0.36 0.66 ± ± ± ± 4.41 5.51 1.58 ± 3.91 ± 4.381.97 1.02 ± 1.45 1.08 ± 1.52 0.00 0.47 0.00 0.00 0.00 0.63 ± 1.71 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.09 0.00 0.00 0.34 ± 0.84 0.07 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 2.02 ± 9.49 Tetraphyllin B 0.27 0.12 0.73 ± ± ± 4.97 ± 0.454.91 ± 0.591.13 ± 1.61 1.14 1.40 ± 1.38 0.39 ± 1.49 0.03 0.12 ± 0.20 0.00 0.00 0.00 ± 0.270.13 0.24 ± 0.48 0.95 0.09 ± 0.28 0.00 0.09 0.00 0.00 0.74 0.00 0.38 ± 0.83 0.30 ± 0.75 0.00 0.00 0.81 0.00 ± 0.250.11 ± 7.58 3.74 3.50 ± 3.94 2.57 ± 3.13 ± 11.81 10.74 Epilotaustralin ± 6.11 7.93 ± 2.71 7.15 ± 8.58 7.73 ± 0.689.58 5.50 ± 7.36 5.02 ± 2.66 ± 8.805.10 5.48 1.22 ± 0.91 1.57 ± 1.19 ± 0.280.11 0.55 ± 0.95 0.08 ± 0.45 0.79 ± 0.56 0.81 ± 1.60 3.58 ± 11.12 3.07 ± 4.35 3.88 ± 2.37 6.40 2.98 2.51 ± 4.58 2.77 ± 3.90 2.20 ± 1.83 2.08 ± 0.16 2.98 ± 3.62 8.39 ± 5.53 41.21 ± 5.29 10.55 ± 2.30 45.78 ± 24.24 12.20 ± 10.18 12.89 12.85 ± 14.28 Lotaustralin 7.64 3.57 1.94 16.43 ± ± ± ± ± 10.32 7.47 ± 3.32 7.96 9.23 1.87 ± 1.66 0.40 ± 1.15 0.45 ± 0.77 ± 0.41 0.19 3.77 ± 10.68 3.93 6.84 ± 9.67 2.57 ± 5.17 17.28 ± 8.43 17.91 19.16 27.56 ± 8.33 37.51 24.37 ± 8.28 ± 13.0845.18 16.65 ± 2.41 13.28 ± 7.08 15.42 26.30 ± 10.38 25.44 ± 7.88 12.50 ± 7.63 12.45 ± 33.8158.11 33.02 ± 6.55 54.18 ± 31.07 30.66 38.15 ± 1.47 38.82 ± 9.22 Linamarin 43.38 ± 4.58 Heliconius erato Heliconius Agraulis vanillae Agraulis Species luciana vanillae Agraulis vanillae vanillae Agraulis huascuma juno Dione miraculosa juno Dione isabellaEueides dissoluta hippolinus isabella Eueides cupidineus aoede Heliconius jamesi burneyi Heliconius vazquezae charithonia Heliconius congener congener Heliconius joroni demeter Heliconius doris doris Heliconius viridis doris Heliconius primularis eleuchia Heliconius cyrbia erato Heliconius demophoon erato Heliconius emma erato Heliconius favorinus erato Heliconius ucayalensis eratosignis Heliconius aerotome ethilla Heliconius Eueides aliphera aliphera Dryas iulia moderata Dryadula phaetusa Dryadula Eueides lybia lybia lybia Eueides Eueides lampeto acacetes Eueides isabella eva Dione juno Eueides isabella Heliconius doris Heliconius Eueides tales calathus TABLE 2 MeanTABLE concentration and associated standard deviation for each compound detected 2684 | SCULFORT et al. (Continues) 1.3 ± 1.8 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.56 ± 1.33 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.46 ± 1.25 0.00 0.00 0.00 0.00 0.38 ± 1.48 Dihydrogynocardin ± 3.442.74 1.80 ± 1.60 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.72 ± 1.45 0.00 0.00 0.00 0.81 ± 2.71 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.92 ± 2.87 0.00 0.00 0.00 0.00 ± 0.540.10 Gynocardin 2.22 ± 3.28 0.00 Tetraphyllin A stereoisomer 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.16 ± 1.55 ± 9.95 ± 10.581.76 1.75 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.09 0.00 0.00 0.00 ± 0.440.10 0.00 0.00 ± 0.490.13 Tetraphyllin A ± 9.02 4.76 1.09 ± 3.79 1.24 ± 2.77 0.00 0.00 0.00 0.50 ± 1.22 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.33 ± 0.99 0.00 0.00 0.00 0.55 ± 2.20 0.00 0.39 ± 1.87 ± 7.20 2.96 14.80 ± 25.64 ± 40.8775.92 28.91 ± 4.55 33.20 ± 37.86 38.17 ± 40.18 Epivolkenin 0.10 ± 4.57 4.03 ± 4.27 1.07 ± 6.58 0.00 0.24 ± 0.88 0.00 0.00 0.00 0.00 0.00 0.35 ± 0.86 0.00 0.03 0.77 ± 1.06 0.00 0.00 0.00 ± 6.19 0.94 0.00 0.25 ± 0.43 0.61 0.00 0.00 0.00 0.23 ± 0.70 0.00 0.00 0.15 ± 0.65 3.78 ± 5.10 3.31 ± 4.66 Tetraphyllin B 5.29 1.75 1.88 ± 0.89 ± 1.871.13 1.49 ± 3.00 1.53 ± 1.82 1.43 ± 3.17 0.00 0.58 ± 0.86 0.67 ± 0.96 0.47 ± 0.81 0.00 ± 0.650.71 0.08 ± 0.26 ± 1.280.91 0.00 0.00 0.00 0.42 ± 0.63 ± 1.370.97 0.34 ± 0.51 0.57 ± 0.53 ± 1.10 0.74 3.58 ± 0.73 ± 3.93 3.74 3.98 ± 2.72 2.19 2.79 ± 3.17 2.30 ± 1.75 2.45 ± 2.98 Epilotaustralin 0.16 2.26 1.97 ± ± ± ± 0.68 7.69 ± 8.42 7.61 ± 2.68 9.16 ± 11.69 9.79 5.27 5.56 ± 3.31 5.23 ± 4.55 5.25 ± 2.47 5.02 ± 3.93 5.27 5.38 ± 3.18 5.84 ± 0.28 5.32 ± 3.41 ± 3.365.76 5.29 ± 3.75 4.92 ± 3.42 1.92 1.66 ± 1.95 0.09 3.26 6.37 ± 2.44 6.31 6.15 ± 6.45 8.07 ± 8.41 11.63 ± 8.28 16.62 ± 8.05 13.16 13.66 ± 8.23 13.09 34.61 ± 15.06 Lotaustralin 7.71 ± 7.82 ± 3.59 9.42 ± 1.85 9.76 0.00 3.44 ± 2.77 6.72 17.13 ± 4.95 17.01 ± 8.8917.80 ± 0.92 17.56 ± 9.55 17.83 ± 7.53 17.88 19.74 ± 15.0037.34 ± 9.00 14.41 ± 8.3314.52 ± 10.41 11.70 24.60 ± 5.99 10.15 ± 5.6610.94 ± 1.11 10.73 10.28 ± 10.41 10.59 ± 10.80 15.82 ± 8.52 15.5 ± 8.49 18.51 ± 11.97 20.59 20.09 ± 8.87 12.26 ± 4.15 12.98 ± 14.65 Linamarin aglaope (hybrid) aglaope Heliconius hecale Heliconius Species felix hecale Heliconius melicerta hecale Heliconius zuleika hecale Heliconius hewitsoni Heliconius melpomeneHeliconius aglaope melpomene Heliconius amaryllis melpomene Heliconius amaryllis melpomene Heliconius rosina melpomene Heliconius arcuella numata Heliconius bicoloratus numata Heliconius lyrcaeus numata Heliconius tarapotensis numata Heliconius zobryssi numata Heliconius pachinus Heliconius pardalinus Heliconius butleri pardalinus Heliconius sergestus pardalinus Heliconius magdalena sara Heliconius sotericus telesiphe Heliconius thelxinoe timareta Heliconius timareta timareta Heliconius flavescens wallacei Heliconius xanthocles Heliconius Heliconius sara sara Heliconius himera Heliconius Heliconius numata Heliconius Heliconius sara Heliconius Heliconius timareta Heliconius TABLE 2 (Continued) TABLE SCULFORT et al. | 2685

32 Pillai342 = 1.735, p < .001) highlight that CG sequestration is widely distributed among the Heliconiini tribe, and may depend on local host plant availability and host plant adaptation. 0.00 0.00 0.00 0.00 0.00 0.00 Dihydrogynocardin 3.2 | Evolution of cyanogenic glucoside profiles in Heliconiini

CG profiles in Heliconiini species (n = 33) displayed a weak but sig- 0.00 0.00 0.00 0.00 0.00 Gynocardin 2.70 ± 0.52 nificant phylogenetic signal (Kmult = 0.311, p = .023). In Heliconius, the largest genus in the Heliconiini radiation, the phylogenetic signal

was also moderate but still significant (Kmult = 0.558, p = .029). In the genus Heliconius, many species have subspecies living in different localities, where individuals display locally mimetic color patterns. To 0.00 Tetraphyllin A stereoisomer 0.00 0.00 0.00 0.00 0.00 test whether the natural selection act on the evolution of defenses due to the evolution of mimetic color pattern, we then estimated the phylogenetic signal in the genus Heliconius at the taxonomic level of subspecies (n = 55). We observed that the phylogenetic signal of 0.27 0.00 0.00 0.00 0.00 0.00 mean CG concentrations then become weaker and nonsignificant Tetraphyllin A

(Kmult = 0.084, p = .055), probably because of important variation among subspecies, consistent with the hypothesis of variations in the strength of selection regarding defenses in different mimicry rings. Intraspecific variations of defenses between localities (four 3 0.00 0.00 0.00 0.00 0.00 Pillai 44.41 Epivolkenin countries, MANOVA, 371 = 0.546, p < .001) could then be ex- plained by either (a) variation in the mimetic community abundance and levels of defenses in comimetic species or (b) variation in host plant availability or host plant specialization levels. 0.75 0.00 0.00 0.00 0.00 0.00 Tetraphyllin B 3.3 | Ecological factors influencing the evolution of cyanogenic glucoside profiles in Heliconiini

To explore the contribution of shared ancestry on one hand, and of eco- 0.00 0.00 0.00 0.00 0.00 0.00 Epilotaustralin logical factors influencing the evolution of defenses on CG variation on the other hand, we drew a phylochemospace displaying average chemi- cal profile of the different subspecies (Figure 3). We observed that sub- species belonging to distinct mimicry rings sometimes had very distinct ± 5.12 9.23 1.34 ± 0.79 0.00 3.33 2.68 ± 0.93 2.03 chemical profiles, for example H. erato favorinus (n = 31), H. erato emma Lotaustralin (n = 5), H. erato demophoon (n = 3), and H. erato cyrbia (n = 1) (MANOVA, 3 Pillai36 = 2.002, p < .001). The distantly related comimics H. melpomene rosina (n = 4) and H. erato demophoon (n = 3) are located closely on the phylochemospace (Figure 3), because of their similar chemical profiles 1 (MANOVA, Pillai = 0 615, p = .621). Similarly, H. melpomene amaryllis ± 3.547.87 . ± 0.349.85 5 5.37 0.00 ± 3.41 17.55 10.38 Linamarin (n = 21) and its comimic H. erato favorinus (n = 31) are located closely in the phylochemospace but their CG profiles were still significantly dif- 1 ferent (MANOVA, Pillai50 = 0.759, p < .001). Overall, the mimicry ring was significantly associated with CG profiles, suggesting that individuals from different species belonging to the same mimicry ring had similar chemical defenses (Table 3). Nevertheless, this association was no longer significant when con- trolling for shared ancestry, suggesting that the similarity in defense levels could be mainly due to increased phylogenetic proximity Species melior xanthocles Heliconius zamora xanthocles Heliconius diatonica Philaethria dido dido Philaethria panamensis dido Philaethria Philaethria dido Philaethria TABLE 2 (Continued) TABLE Note: present We data for both species and subspecies. CG concentrations are given in µg/mg of dried body mass. within mimicry rings (Table 3). 2686 | SCULFORT et al.

r n

B A s n n reoisome e Log10 [CG] µg/mg DW rin

TOTAL Lin hydrogynocardi et. A st TOTAL CG Linama Lotaustralin EpilotaustralinTetraphyllin Epivolkeni Tetraphyllin T Gynocardi Di 100 58 Heliconius hecale zuleika Heliconius hecale melicerta Heliconius hecale felix 10 Heliconius pardalinus sergestus Heliconius pardalinus butleri Heliconius ethilla aerotome 1 Heliconius numata zobryssi Heliconius numata tarapotensis 10 Heliconius numata lyrcaeus Heliconius numata bicoloratus Heliconius numata arcuella LotEpilo Heliconius melpomene rosina Heliconius melpomene amaryllis aglaope 46 10 Heliconius melpomene amaryllis Heliconius melpomene aglaope 10 Heliconius timareta timareta 1 Heliconius timareta thelxinoe Heliconius pachinus 1 Heliconius burneyi jamesi Heliconius wallacei flavescens 0.1 Non-pupal-mating Heliconius doris viridis Heliconius doris doris Heliconius xanthocles zamora TetB Epivo Heliconius xanthocles melior Heliconius aoede cupidineus 575 Heliconius congener congener Heliconius eleuchia primularis 10 Heliconius Heliconius hewitsoni 1 Heliconius sara sara Heliconius sara magdalena 1 Heliconius charithonia vazquezae Heliconius demeter joroni 0.1 0.1 Heliconius eratosignis ucayalensis Heliconius erato favorinus Pupal-mating Heliconius erato emma TetA TetA ste. Heliconius erato demophoon 4 Heliconius erato cyrbia 10 Heliconius himera Heliconius telesiphe sotericus Eueides isabella hippolinus Eueides isabella eva 1 Eueides isabella dissoluta Eueides lampeto acacetes Eueides tales calathus 0.1 1 Eueides lybia lybia Eueides aliphera aliphera GynDihy Dione juno miraculosa 33 Dione juno huascuma Agraulis vanillae vanillae 1 Agraulis vanillae luciana 1 Heliconiini Philaethria dido panamensis Philaethria dido dido Philaethria diatonica Dryas iulia moderata Dryadula phaetusa 0.1

FIGURE 2 Qualitative and quantitative variations for the nine studied CGs across Heliconiini subspecies. Phylogenetic tree is adapted from (Kozak et al., 2015). The left column represents the total CG mean concentration (n = 375 individuals in 55 subspecies). Following column presents the average of each CG concentration. Concentrations are in µg of CG per mg of dried weigh (body + wings) in a logarithmic scale. A black box signifies either the absence of the CG or insufficient data for measurement. A colored filled box indicates that the corresponding CG has been reported in at least one individual of the species. Color gradient is from white corresponding to the minimum reported concentration to the darkest color corresponding to the maximal reported concentration

The level of host plant specialization could also influence the butterfly ability to choose and survive on different plants but also evolution of defense in Heliconiini. Indeed, we noticed that the on the local host plant availability). chemical profiles of butterflies depended on their level of host plant specialization, although this effect is mostly driven by phyloge- netic proximity (Table 3). Because there is substantial geographical 3.4 | Geographical variation in chemical profiles variation in the level of specialization, we also compared chemical defenses among subspecies: individuals from host plant-specialist In general, variation in CGs was lower within than between mimicry subspecies were generally more chemically defended (mean total rings (Table 3). Mimicry rings are composed of different species found [CGs] = 39.2 µg/mg DW) than generalist (26.5 µg/mg DW; Table 3, in sympatry, they can therefore differ in local abundance but also in Figure 4). Specialist subspecies sequestered more CGs (19.2 µg/ host plants availability. Mimetic communities exhibiting the same color mg DW) than generalist subspecies (3.8 µg/mg DW; ANOVA, pattern (e.g., postman color pattern, Figure 5) are composed of similar F1 373 = 53.01, p < .001). This is pointing at the effect of host plant species, but still display strikingly different chemical profiles (Figures 5 specialization on chemical profiles that could substantially vary and 6). Both color pattern and locality indeed have a significant asso- among localities (note that such specialization could depend on the ciation with chemical profiles, as well as the interaction between these SCULFORT et al. | 2687

Lotaustralin Dihydrogynocardin Linamarin Epilotaustralin Epivolkenin

H. charithonia vazquezae Tetraphyllin A Gynocardin Tet. A. stereoisomer H. erato demophoon H. melpomene rosina Tetraphyllin B

A. vanillae vanillae

E. isabella dissoluta E. isabella eva

PC2 (38.75%) E. lybia lybia H. erato cyrbia 0246 E. lampeto acacetes H. numata lyrcaeus H. erato emma H. eleuchia primularis H. melpomene amaryllis −2 H. erato favorinus H. congener congener H. eratosignis ucayalensis

−6 −4 −2 0 2468 PC1 (52.14%) Blue Green Postman Panama Postman reverse Tiger Mimicry ring Dennis ray Orange Postman Ecuador/Peru Rayed yellow Other

FIGURE 3 Phylochemospace depicting the relationships between phylogenetic history and the mean CG concentration in Heliconiini subspecies. Visualization in two dimensions of the distribution of the variation in CG profiles. Dark line represents the phylogenetic tree modified from Kozak et al., (2015) to plot subspecies used in our analyses (n = 55 subspecies). Dots are mean imputed CG profile per subspecies. Color indicates the mimicry ring subspecies belong to (Table 1). Heliconius erato subspecies from distinct mimicry rings also differ in their mean chemical profiles (H. e. cyrbia in the “Other” mimicry ring from Ecuador, H. e. emma from Dennis-ray ring from Peru, H. e. favorinus from Postman ring from Peru and H. e. demophoon from Postman ring from Panama). H. erato and H. melpomene subspecies have increased size dot and are illustrated by a photo

TABLE 3 Comparisons of CG profile (MANOVA) between and 4 | DISCUSSION among mimicry rings and host plant specialization levels MANOVA on mean Regular Phylogenetic 4.1 | Phylogenetic history partly explains the per subspecies Mimicry ringa distribution of CGs across Heliconiini species (n = 55) 9 9 Pillai36 = 2.736, p < .001 Pillai36 = 2.736, p < .582 We observed that mimicry rings had different levels of CG profiles and Host plant specializationa total concentrations, but these differences are mostly driven by close 1 1 phylogenetic relatedness among mimetic species. Our results in wild- Pillai53 = 0.446, Pillai53 = 0.446, p < .001 p < 1.000 caught individuals are thus consistent with the significant phyloge- MANOVA on Regular netic signal in CG profile observed in captive-bred Heliconiini (Castro, a Zagrobelny, et al., 2019). Nevertheless, the phylogenetic signal associ- interindividual Mimicry ring variation (n = 375) 10 ated with CG profile is stronger when considering species rather than Pillai364 = 1.209, p < .001 subspecies, suggesting that despite a strong effect of the divergence Host plant specializationa between clades (ancient node), substantial variation within species is Pillai1 373 = 0.165, p < .001 observed in our wild-caught individuals, probably driven by ecological Note:: To compare the effect of mimicry rings and host plant factors acting on the different mimetic subspecies. specialization on CG profiles with phylogenetic effect, we performed a MANOVA using the mean concentration per subspecies (n = 55 subspecies). Then MANOVA was performed on CG profiles using the whole dataset to test for interindividual variation (n = 375 individuals), 4.2 | Geographic variation in mimicry rings impacts without testing the effect of phylogeny. CG profiles aNote that each factor was tested using an independent MANOVA. The important variation in CG profile observed within species is mostly two factors, even when controlling for the species effect (Table 4). This explained by variations between subspecies living in different geo- suggests that geographical variations in local abundances of mimetic graphic range. For instance, Panamanian subspecies of A. vanillae and patterns and/or in local host plants availability and specialization levels H. erato were more chemically defended than Southern subspecies of may influence the defenses of Heliconiini butterflies. the same two species. Subspecies generally differ in wing color pattern 2688 | SCULFORT et al.

highlights that host plant interaction and geography are important eco- * logical factors shaping variations in chemical defenses within species. *

4.3 | How host plant specialization shapes 50 chemical defenses CGs Total Synthesized Indeed, host plant range and preference vary locally in some spe- Sequestered [CG] µg/mg DW cies (Smiley, 1978), so that variation in putatively-sequestrated

0 CGs in butterflies probably reflects host plant availability and use across sampled localities. For example, H. melpomene has a Generalist Specialist wider range of host plant species in its eastern distribution area. FIGURE 4 Amount of chemical defenses according to host In Central America, it feeds on P. menispermifolia or P. oerstedii de- plant specialization. CG concentrations are given in µg/mg of dried pending on the localities but feeds preferentially on P. platyloba in body mass. We pooled generalist subspecies (n = 210 individuals Peru, (Billington, Thomas, & Gilbert, 1990; Jiggins, 2016). This em- distributed in 32 subspecies) on the left and specialist subspecies phasizes the plasticity in the host plant range of many Heliconiini (n = 165 individuals distributed in 23 subspecies) on the right. We represented the total amount of CG (red boxplot) that sums species and the importance of local adaptation with Passiflora spe- synthesized (green boxplot) and sequestered (blue boxplot) CG cies. Local patterns in host plant use by Heliconiini are likely re- concentrations. Asterix shows significant statistical difference flected in their CG profile. The binary generalist/specialist classification used here is a and geographic distribution, pointing at the influence of ecological fac- rough simplification of the host plant specialization spectrum. tors in shaping the variation in CG concentration profile in Heliconiini. Nevertheless, we still observed, as expected, that specialist sub- Although Heliconius species from the pupal-mating and nonpupal- species had higher concentrations of putatively-sequestrated CGs mating clades are phylogenetically distant, they can be involved in the (Engler & Gilbert, 2007; Jiggins, 2016). However, we did not detect same mimicry ring. This is the case for H. erato demophoon and H. mel- any correlation between the level of host plant specialization and pomene rosina, which are part of the postman Panama mimicry ring the synthesis/sequestration balance, contrary to previous studies and presented similar CG profiles, suggesting either an effect of the where synthesis and sequestration were shown to be negatively cor- mimetic interactions and/or of the similarity in local host plant chem- related traits, with fluctuant intensity across the phylogeny (Castro, istry. By sampling wild butterflies from different countries, our study Zagrobelny, et al., 2019; Engler & Gilbert, 2007).

Synthesized CGs Sequestered CGs Linamarin Tetraphyllin B Gynocardin Lotaustralin Epivolkenin Dihydrogynocardin Epilotaustralin Tetraphyllin A

100

FIGURE 5 Variation in chemical profiles of individuals from the nine studied mimicry rings, located in different regions of Central and South America.

/mg DW 50 CG concentrations are given in µg/mg DW. Mimicry rings from left to right,

[CG] µg with illustrations of the color pattern: blue (6 subspecies, n = 66 individuals), Dennis ray (10 subspecies, n = 39), green (3 subspecies, n = 4), orange (8 0 subspecies, n = 73), postman Panama (2 subspecies, n = 7), postman reverse (2 subspecies, n = 6), postman from Ecuador Brazil Ecuador Panama Brazil Panama EcuadorEcuador Panama Brazil and Peru (5 subspecies, n = 57), rayed Ecuador Peru Peru Ecuador Peru Peru yellow (2 subspecies, n = 7), and tiger (11 Panama Panama subspecies, n = 78). White boxplots are Peru Peru mean total CG concentration SCULFORT et al. | 2689

Mimicry ring

100 Blue Dennis ray Green Orange

Postman Panama 50 Postman Reverse Postman South Rayed yellow Tiger

Other 0

i a a a a a s s s n a a e e a s s a e a a is ta llis us ev idis elix ell nus c inus yssi ry atus i r ri aret a osina or r br butler amor witsoni m vanillae v escens z oderata ris doriis vir imular zquezae yalensis zo fa aglaop m av eyi jamesi va ia diatonicaia dido dido rn yllis r h nus sergestura magdalen anillae bu vanillae luciana ethr yadula phaetus numata lyrcaeu allacei fl yas iulia Heliconius himer s numata pardalinus Dr Eueides lybia lybia rithonia a dido panamensis e amar w Heliconius pach ri Dr Eueides isabella Heliconius he n numata tarapotensis Heliconius sara sara PhilaetPhila Dione juno huascum Heliconius do Dione juno miraculosa Eueides tales calathu Heliconius dor Heliconius hecale f Heliconius eratoHeliconius cyrbia erato emma Agraulis onius timareta thelxino Agraulis v c onius xanthocles Eueides aliphera alipher Heliconius hecale zuleik Eueides isabella dissolut Heliconius Heliconius demeter joroni Heliconius erato Eueides lampeto acacete Heliconius ethilla aerotome Heliconius numata arcu onius pardali Eueides isabella hippolinus Heliconius hecale melicer Heliconius Heliconiu Heliconius sa Heliconiusc Heliconius timareta ti Philaeth Heliconius aoede cupidineus Heliconius erato demophoo Heli Heliconius xanthocles melior Heliconius melpomene Heliconius telesiphe sote Heliconius eleuchia pr Heliconius numataeliconius bicolo Heliconius Helic Heliconius congener congener Heliconius melpomene aglaop H Heli Heliconius melpomene am Heliconius cha Heliconius eratosignis uca

Heliconius melpome

FIGURE 6 Total CG concentration per subspecies. Concentrations are given in µg/mg DW. Boxplot colors correspond to the associated mimicry ring with legend on the right. Subspecies are listed in alphabetical order from left to right (n = 55 subspecies)

TABLE 4 Variation of CG chemical profile between individuals (n = 375)

Regular MANOVA on interindividual variation (n = 375)

Degrees of freedom of Degrees of freedom of p-value associated with df Pillai F statistic the numerator the denominator the F statistic F9 Colour pattern 9 1.455 325 = 6.965 81 2,925 .001 F3 Locality 3 1.167 325 = 22.544 27 957 .001 F8 Colour pattern + 29 0.540 325 = 2.607 72 2,592 .001 Locality F28 Species 8 2.371 325 = 4.153 252 2,925 .001 F1 Specialization 1 0.247 325 = 11.546 9 317 .001 Note: MANOVA tests if there is difference for the CG chemical profiles between groups (listed in left column). Residuals = 325.

As CGs are Passiflora secondary metabolites, their production 4.4 | Variability of CG profiles within mimicry may vary in space, time and across tissues depending on abiotic and rings and Müllerian mimicry biotic conditions exert on plant. Thus, reported putatively-seques- trated CGs in our study on wild butterflies are potentially a subset of Variation in CG concentrations between mimicry rings observed here the CGs contained in locally available Passiflora host plants. The evo- had already been reported in a study based on colorimetric assays (to lution of Heliconiini chemical defense profile would thus be shaped investigate total CG concentration per individual regardless of each CG by both host plant specialization of the different butterfly species identity) (Arias et al., 2016). This effect of mimicry on the individuals be- and available Passiflora host plants variations across the geograph- longing to different co-occurring mimicry rings is thus not necessarily ical areas. equally defended, and potentially perceived with different degrees of 2690 | SCULFORT et al. aversion by predators. Recently, an experiment using domestic chicks ACKNOWLEDGMENTS shows that beyond a certain CG concentration, birds learned to avoid This work was supported by a grant from Agence Nationale de la the prey at a similar speed (Chouteau, Dezeure, Sherratt, Llaurens, & Recherche under the LabEx ANR-10-LABX-0003-BCDiv, in the pro- Joron, 2019). Variations in the level of CGs observed within and among gram “Investissements d'avenir” number ANR-11-IDEX-0004-02 mimicry rings might thus not directly translate into variation in learning attributed to OS and Paris city council grant Emergence to VL. EC behavior by predators, so that the evolution of high chemical defense would like to thank Professor Chris Jiggins for the financial sup- in some Heliconiini would not necessarily be promoted by natural se- port through the European Research Council grant number 339873 lection exerted by predators in mimetic prey. Furthermore, it is cur- (Acronym: SpeciationGenetics) and the Marie Curie Actions for her rently unknown whether predator rejection behavior depends on the fellowship (Acronym: Cyanide Evolution). total concentration of CG or is mostly shaped by the presence of key We would like to thank Gerardo Lamas from the Natural History CGs with a particularly repellent taste. Chemical defenses are also a Museum of Lima (Peru) and Peruvian authorities (Servicio Nacional complex cocktail (Speed, Ruxton, Mappes, & Sherratt, 2012) with com- Forestal y de Fauna Silvestre) for delivering research and export per- ponents acting through synergetic or antagonist effects. mits (Research permit number: RDG 0373-2017-SERFOR-DGGSPFFS). Predator communities and strength in predation pressure act- We acknowledge the collecting permits issued by the Ministry of the ing on aposematic prey vary in space and time, as demonstrated in Environment of Ecuador (MAE-DNB-CM-2016-0045), the Ministry of the field using artificial poison frogs and caterpillars (Chouteau & the Environment of Panama (SE/AP-11-17) and Brazil (TBC). Collection Angers, 2011; Mappes, Kokko, Ojala, & Lindström, 2014). Predator of new Brazilian samples was made under SISBIO license 59194-1. sensibility to detect bitterness of CGs and to endure unpleasant The export of new material from Brazil was made through the Federal taste varies (Li & Zhang, 2014), as well as their tolerance toward University of Pará under general approval for export of biological mate- cyanide (Cardoso, 2019). Indeed, based on how hungry they are, rial by the Council for Management of Genetic Patrimony (Conselho de avian predators may decide to feed on unpalatable butterflies Gestão de Patrimônio Genético), Deliberation No. 19 of the 19th March, (Chai, 1986; Marshall, 1908). The geographic variation in chemical 2003, in reference to process number 02018.005329/02-10(73), along profile detected here might therefore be influenced by both host with the specific documentation required by that approval for each in- plant availability and composition of predator communities. But the dividual exportation event. strong phylogenetic signal detected on CG profiles, and the high O.S wishes to thank Melanie McClure, Mathieu Chouteau, sensitivity of predator to CG suggests that the evolution of elevated Marion Cau, Mario Tuanama, and Ronald Mori-Pezo for precious levels of chemical defense is not directly related to color pattern support on the field. O.S thanks David Ian Pattison for technical as- evolution. sistance with LC-MS/MS, and Charline Pinna, Camille Le Roy and Léa Terray for help on R coding. We thank Lawrence E. Gilbert and Neil Rosser for their expertize on Heliconiini host plant specialization. 5 | CONCLUSIONS We thank Richard Merrill for inspiring discussions about unpalatably variation across localities. Our study sheds light on the evolution of CGs in Heliconiini butterflies, and highlights the strong effect of evolutionary history in the varia- CONFLICT OF INTEREST tion of CG profile observed between species. Variation in CG profiles None declared. between mimicry rings seems to be mostly driven by phylogenetic relatedness between mimetic species. Nevertheless, the strong varia- AUTHOR CONTRIBUTIONS tion observed between individuals belonging to different mimicry rings V.L, B.N, O.S, and M.E. conceived the study. O.S, K.M.K, and V.L. col- within species suggests that other ecological factors might be at play. lected the specimens. S.B welcomed O.S at the Department of Plant Some species seem to rely on de novo synthesis only, whereas other and Environmental Sciences, University of Copenhagen, Denmark, species mostly perform CG sequestration from Passiflora host plants. so she could performed the chemical analyses with help from E.C, Many species rely on a combination of these two pathways for CG ac- O.S. performed statistical analyses and wrote the manuscript with quisition, which contributes to substantial variation of chemical profiles contributions from all authors. All authors participated in construc- both between species and among species. Geographic variation in host tive discussions and approved manuscript final version. plants, but also abundance of mimicry rings could also influence the CG profile: The individual predation risk is indeed lower in abundant DATA AVAILABILITY STATEMENT mimicry rings as compared with rare ones (Chouteau, Arias, & Joron, Raw data file describing each compound and concentration per 2016), so that selection for higher distastefulness might be higher in lo- individual is available on Dryad following the link: https​://doi. calities where a given mimicry ring is at low density. Ecological studies org/10.5061/dryad.ghx3f​fbjt estimating local host plant and predator community variations, as well as local abundances of mimetic communities, would now be required ORCID to better understand the selective pressures shaping chemical defense Ombeline Sculfort https://orcid.org/0000-0002-1583-824X evolution in mimetic species. Erika C. P. de Castro https://orcid.org/0000-0002-4731-3835 SCULFORT et al. | 2691

REFERENCES Chassagne, D., Crouzet, J. C., Bayonove, C. L., & Baumes, R. L. (1996). Identification and quantification of passion fruit cyanogenic glyco- Adams, D. C. (2014). A generalized K statistic for estimating phyloge- sides. Journal of Agriculture and Food Chemistry, 44, 3817–3820. https​ netic signal from shape and other high-dimensional multivariate ://doi.org/10.1021/jf960​381t data. Systematic Biology, 63, 685–697. https​://doi.org/10.1093/sysbi​ Chouteau, M., & Angers, B. (2011). The role of predators in maintain- o/syu030 ing the geographic organization of aposematic signals. The American Alcock, J. (1970a). Punishment levels and the response of black- Naturalist, 178, 810–817. https​://doi.org/10.1086/662667 capped chickadees (Parus atricapillus) to three kinds of ar- Chouteau, M., Arias, M., & Joron, M. (2016). Warning signals are under tificial seeds. Behaviour, 18, 592–599. https​://doi. positive frequency-dependent selection in nature. PNAS, 113, 2164– org/10.1016/0003-3472(70)90057-6 2169. https​://doi.org/10.1073/pnas.15192​16113​ Alcock, J. (1970b). Punishment levels and the response of white- Chouteau, M., Dezeure, J., Sherratt, T. N., Llaurens, V., & Joron, M. throated sparrows (Zonotrichia albicollis) to three kinds of artificial (2019). Similar predator aversion for natural prey with diverse tox- models and mimics. Animal Behaviour, 18, 733–739. https​://doi. icity levels. Animal Behaviour, 153, 49–59. https​://doi.org/10.1016/j. org/10.1016/0003-3472(70)90019-9 anbeh​av.2019.04.017 Arias, M., Meichanetzoglou, A., Elias, M., Rosser, N., de-Silva, D. L., Nay, Conn, E. E. (1980). Cyanogenic compounds. Annual Review of Plant Physiology, B., & Llaurens, V. (2016). Variation in cyanogenic compounds con- 31, 433–451. https​://doi.org/10.1146/annur​ev.pp.31.060180.002245 centration within a Heliconius butterfly community: Does mimicry Cortesi, F., & Cheney, K. L. (2010). Conspicuousness is correlated with explain everything? BMC Evolutionary Biology, 16, 272. https​://doi. toxicity in marine opisthobranchs. Journal of Evolutionary Biology, 23, org/10.1186/s12862-016-0843-5 1509–1518. https​://doi.org/10.1111/j.1420-9101.2010.02018.x Benson, W. W., Brown, K. S., & Gilbert, L. E. (1975). Coevolution of plants de Castro, É. C. P., Demirtas, R., Orteu, A., Olsen, C. E., Motawie, M. S., and herbivores: Passion flower butterflies. Evolution, 29, 659–680. Cardoso, M. Z., … Bak, S. (2019). The dynamics of cyanide defences in https​://doi.org/10.1111/j.1558-5646.1975.tb008​61.x the life cycle of an aposematic butterfly: Biosynthesis versus seques- Bezzerides, A. L., McGraw, K. J., Parker, R. S., & Husseini, J. (2007). Elytra tration. Biochemistry and Molecular Biology, 116, 103259. https​ color as a signal of chemical defense in the Asian ladybird beetle ://doi.org/10.1016/j.ibmb.2019.103259 Harmonia axyridis. Behavioral Ecology and Sociobiology, 61, 1401– de Castro, É. C. P., Zagrobelny, M., Zurano, J. P., Cardoso, M. Z., Feyereisen, 1408. https​://doi.org/10.1007/s00265-007-0371-9 R., & Bak, S. (2019). Sequestration and biosynthesis of cyanogenic glu- Billington, H. L., Thomas, C. D., & Gilbert, L. E. (1990). Variation in cosides in passion vine butterflies and consequences for the diversi- stage-specific mortality patterns of a specialist herbivore on dif- fication of their host plants. Ecology and Evolution, 9(9), 5079–5093. ferent host plant clones. Functional Ecology, 4, 721. https​://doi. Engler, H., & Gilbert, L. E. (2007). De novo synthesis vs. sequestration: org/10.2307/2389438 Negatively correlated metabolic traits and the evolution of host plant Blomberg, S. P., Garland, T., & Ives, A. R. (2003). Testing for phyloge- specialization in cyanogenic butterflies. Journal of Chemical Ecology, netic signal in comparative data: Behavioral traits are more labile. 33, 25–42. https​://doi.org/10.1007/s10886-006-9207-8 Evolution, 57, 717–745. https​://doi.org/10.1111/j.0014-3820.2003. Engler, H., Spencer, K. C., & Gilbert, L. E. (2000). Insect metabolism: t b 0 0 2 8​ 5 . x Preventing cyanide release from leaves. Nature, 406, 144–145. https​ Blount, J. D., Speed, M. P., Ruxton, G. D., & Stephens, P. A. (2009). Warning ://doi.org/10.1038/35018159 displays may function as honest signals of toxicity. Proceedings of Goodale, M. A., & Sneddon, I. (1977). The effect of distastefulness of the the Royal Society B: Biological Sciences, 276, 871–877. https​://doi. model on the predation of artificial batesian mimics. Animal Behaviour, org/10.1098/rspb.2008.1407 25, 660–665. https​://doi.org/10.1016/0003-3472(77)90117-8 Boyden, T. C. (1976). Butterfly palatability and mimicry: experi- Harmon, L. J., Weir, J. T., Brock, C. D., Glor, R. E., & Challenger, W. (2008). ments with ameiva lizards. Evolution, 30, 73–81. https​://doi. GEIGER: Investigating evolutionary radiations. Bioinformatics, 24, org/10.1111/j.1558-5646.1976.tb008​82.x 129–131. https​://doi.org/10.1093/bioin​forma​tics/btm538 Braby, M. F., Eastwood, R., & Murray, N. (2012). The subspecies concept Hervé, M. (2019). RVAideMemoire: Testing and plotting procedures for in butterflies: Has its application in taxonomy and conservation biol- biostatistics. ogy outlived its usefulness? Biological Journal of the Linnean Society, Ihalainen, E., Lindström, L., & Mappes, J. (2007). Investigating 106, 699–716. https​://doi.org/10.1111/j.1095-8312.2012.01909.x Müllerian mimicry: Predator learning and variation in prey de- Briolat, E. S., Zagrobelny, M., Olsen, C. E., Blount, J. D., & Stevens, M. fences. Journal of Evolutionary Biology, 20, 780–791. https​://doi. (2019). No evidence of quantitative signal honesty across species org/10.1111/j.1420-9101.2006.01234.x of aposematic burnet moths (: Zygaenidae). Journal of Jaroszewski, J., Olafsdottir, E., Wellendorph, P., Christensen, J., Franzyk, Evolutionary Biology, 32, 31–48. https​://doi.org/10.1111/jeb.13389​ H., Somanadhan, B., … Clausen, V. (2002). Cyanohydrin glycosides Brower, L. P., McEvoy, P. B., Williamson, K. L., & Flannery, M. A. (1972). of Passiflora: Distribution pattern, a saturated cyclopentane de- Variation in cardiac glycoside content of monarch butterflies from rivative from P. guatemalensis, and formation of pseudocyanogenic natural populations in eastern North America. Science, 177, 426–429. -hydroxyamides as isolation artefacts. Phytochemistry, 59, 501–511. https​://doi.org/10.1126/scien​ce.177.4047.426 α https​://doi.org/10.1016/S0031-9422(01)00485-X Brown, K. S. (1981). The biology of heliconius and related genera. Annual Jiggins, C. D. (2016). The ecology and evolution of Heliconius butterflies. Review of Entomology, 26, 427–457. https​://doi.org/10.1146/annur​ Oxford, UK: Oxford University Press. ev.en.26.010181.002235 Jones, P. L., Petschenka, G., Flacht, L., & Agrawal, A. A. (2019). Cardenolide Cardoso, M. Z. (2019). The effect of insect cyanoglucosides on predation intake, sequestration, and excretion by the monarch butterfly along by domestic chicks. BioRxiv. gradients of plant toxicity and larval ontogeny. Journal of Chemical Chai, P. (1986). Field observations and feeding experiments on the re- Ecology, 45(3), 264–277 https​://doi.org/10.1007/s10886-019-01055-7 sponses of rufous-tailed jacamars (Galbula ruficauda) to free-flying but- Joron, M., & Iwasa, Y. (2005). The evolution of a Müllerian mimic in a terflies in a tropical rainforest. Biological Journal of the Linnean Society, spatially distributed community. Journal of Theoretical Biology, 237, 29, 161–189. https​://doi.org/10.1111/j.1095-8312.1986.tb017​72.x 87–103. https​://doi.org/10.1016/j.jtbi.2005.04.005 Chai, P. (1996). Butterfly visual characteristics and ontog- Joron, M., & Mallet, J. L. B. (1998). Diversity in mimicry: Paradox or eny of responses to butterflies by a specialized tropical bird. paradigm? Trends in Ecology & Evolution, 13, 461–466. https​://doi. Biological Journal of the Linnean Society, 59, 37–67. https​://doi. org/10.1016/S0169-5347(98)01483-9 org/10.1111/j.1095-8312.1996.tb014​52.x 2692 | SCULFORT et al.

Josse, J., & Husson, F. (2016). missMDA: A package for handling missing Santos, J. C., & Cannatella, D. C. (2011). Phenotypic integration emerges values in multivariate data analysis. Journal of Statistical Software, 70, from aposematism and scale in poison frogs. PNAS, 108, 6175–6180. 1-31. https​://doi.org/10.1073/pnas.10109​52108​ Kozak, K. M. (2016). Macroevolution and phylogenomics in the adaptive ra- Sherratt, T. N. (2008). The evolution of Müllerian mimicry. diation of Heliconiini butterflies. Thesis, University of Cambridge. Naturwissenschaften, 95, 681. https​://doi.org/10.1007/ Kozak, K. M., Wahlberg, N., Neild, A. F. E., Dasmahapatra, K. K., Mallet, s00114-008-0403-y J., & Jiggins, C. D. (2015). Multilocus species trees show the recent Sidlauskas, B. (2008). Continuous and arrested morphological di- adaptive radiation of the mimetic heliconius butterflies. Systematic versification in sister clades of characiform fishes: A phylo- Biology, 64(3), 505–524. https​://doi.org/10.1093/sysbi​o/syv007 morphospace approach. Evolution, 62, 3135–3156. https​://doi. Le Poul, Y., Whibley, A., Chouteau, M., Prunier, F., Llaurens, V., & Joron, org/10.1111/j.1558-5646.2008.00519.x M. (2014). Evolution of dominance mechanisms at a butterfly Skelhorn, J., & Rowe, C. (2005). Tasting the difference: Do multiple de- mimicry supergene. Nature Communications, 5, 5644. https​://doi. fence chemicals interact in Mullerian mimicry? Proceedings of the org/10.1038/ncomm​s6644​ Royal Society B: Biological Sciences, 272, 339–345. Lê, S., Josse, J., & Husson, F. (2008). FactoMineR: An R package for multi- Smiley, J. (1978). Plant chemistry and the evolution of host specificity: variate analysis. Journal of Statistical Software, 25, 1–18. new evidence from heliconius and passiflora. Science, 201, 745–747. Li, D., & Zhang, J. (2014). Diet shapes the evolution of the vertebrate https​://doi.org/10.1126/scien​ce.201.4357.745 bitter taste receptor gene repertoire. Molecular Biology and Evolution, Speed, M. P. (1993). Muellerian mimicry and the psychology of pre- 31, 303–309. https​://doi.org/10.1093/molbe​v/mst219 dation. Animal Behaviour, 45, 571–580. https​://doi.org/10.1006/ Mappes, J., Kokko, H., Ojala, K., & Lindström, L. (2014). Seasonal changes in anbe.1993.1067 predator community switch the direction of selection for prey defences. Speed, M. P., & Ruxton, G. D. (2007). How bright and how nasty: Nature Communications, 5, 5016. https​://doi.org/10.1038/ncomm​s6016​ Explaining diversity in warning signal strength. Evolution, 61, 623– Marshall, G. A. K. (1908). On Diaposematism, with reference to some 635. https​://doi.org/10.1111/j.1558-5646.2007.00054.x limitations of the Müllerian Hypothesis of Mimicry. Transactions of Speed, M. P., Ruxton, G. D., Mappes, J., & Sherratt, T. N. (2012). the Royal Entomological Society of London, 56, 93–142. https​://doi. Why are defensive toxins so variable? An evolutionary per- org/10.1111/j.1365-2311.1908.tb021​41.x spective. Biological Reviews, 87, 874–884. https​://doi. Merrill, R. M., Dasmahapatra, K. K., Davey, J. W., Dell’Aglio, D. D., org/10.1111/j.1469-185X.2012.00228.x Hanly, J. J., Huber, B., … Martin, S. H. (2015). The diversification of Spencer, K. C., & Seigler, D. S. (1987). Passisuberosin and epipassisub- Heliconius butterflies: What have we learned in 150 years? Journal of erosin: Two cyclopentenoid cyanogenic glycosides from Passiflora Evolutionary Biology, 28, 1417–1438. suberosa. Phytochemistry, 26, 1665–1667. https​://doi.org/10.1016/ Møller, B. L., Olsen, C. E., & Motawia, M. S. (2016). General and stereo- S0031-9422(00)82266-9 controlled approach to the chemical synthesis of naturally occurring Spencer, K. C., Seigler, D. S., & Domingo, J. L. (1983). Tetraphyllins A cyanogenic glucosides. Journal of Natural Products, 79, 1198–1202. and B, deidaclin and epitetraphyllin B from Tetrapathaea tetran- https​://doi.org/10.1021/acs.jnatp​rod.5b01121 dra (Passifloraceae). Phytochemistry, 22, 1815–1816. https​://doi. Müller, F. (1879). Ituna and Thyridia; a remarkable case of mimicry in but- org/10.1016/S0031-9422(00)80277-0 terflies. Kosmos, 100-109. Tober, I., & Conn, E. E. (1985). Cyclopentenylglycine, a precursor of deid- Nahrstedt, A., & Davis, R. H. (1985). Biosynthesis and quantitative relation- aclin in Turnera ulmifolia. Phytochemistry, 24, 1215–1218. https​://doi. ships of the cyanogenic glucosides, linamarin and lotaustralin, in genera org/10.1016/S0031-9422(00)81104-8 of the Heliconiini (Insecta: Lepidoptera). Comparative Biochemistry and Turner, J. R. G. (1967). Some early works on heliconiine butter- Physiology Part B: Comparative Biochemistry, 82, 745–749. https​://doi. flies and their biology (Lepidoptera, Nymphalidae). Journal of org/10.1016/0305-0491(85)90519-X the Linnean Society of London, Zoology, 46, 255–266. https​://doi. Nishida, R. (2002). Sequestration of defensive substances from plants org/10.1111/j.1096-3642.1967.tb005​07.x by lepidoptera. Annual Review of Entomology, 47, 57–92. https​://doi. Wickham, H., Chang, W., Henry, L., Pedersen, T. L., Takahashi, K., Wilke, org/10.1146/annur​ev.ento.47.091201.145121 C., … RStudio (2019). ggplot2: Create Elegant Data Visualisations Using Paradis, E. (2011). Analysis of Phylogenetics and Evolution with R (Springer the Grammar of Graphics. Science & Business Media). Wray, V., Davis, R. H., & Nahrstedt, A. (1983). Biosynthesis of cyano- Pinheiro, C. E. G., & Campos, V. C. (2019). The responses of wild jacamars genic glycosides in butterflies and moths: Incorporation of valine and (Galbula ruficauda, Galbulidae) to aposematic, aposematic and cryp- isoleucine into linamarin and lotaustralin by zygaena and heliconius tic, and cryptic butterflies in central Brazil. Ecological Entomology, species (Lepidoptera). Zeitschrift Für Naturforschung C, 38, 583–588. 44(4), 441–450. https​://doi.org/10.1515/znc-1983-7-815 Prudic, K. L., Timmermann, B. N., Papaj, D. R., Ritland, D. B., & Oliver, J. Yu, G., Smith, D. K., Zhu, H., Guan, Y., & Lam, T.-T.-Y. (2017). ggtree: An r C. (2019). Mimicry in viceroy butterflies is dependent on abundance package for visualization and annotation of phylogenetic trees with of the model queen butterfly. Communications Biology, 2, 68. https​:// their covariates and other associated data. Methods in Ecology and doi.org/10.1038/s42003-019-0303-z Evolution, 8, 28–36. R Core Team (2019a). R: The R Project for Statistical Computing. Vienna, Zagrobelny, M., de Castro, É. C. P., Møller, B. L., & Bak, S. (2018). Austria: R Foundation for Statistical Computing. https​://ww- Cyanogenesis in : from chemical warfare to nuptial gifts. w.R-proje​ct.org/ Insects, 9, 51. https​://doi.org/10.3390/insec​ts902​0051 Revell, L. J. (2012). phytools: An R package for phylogenetic compara- tive biology (and other things). Methods in Ecology and Evolution, 3, 217–223. https​://doi.org/10.1111/j.2041-210X.2011.00169.x How to cite this article: Sculfort O, de Castro ECP, Kozak KM, Rowland, H., Mappes, J., Ruxton, G. D., & Speed, M. P. (2010). Mimicry et al. Variation of chemical compounds in wild Heliconiini between unequally defended prey can be parasitic: Evidence for quasi-Batesian mimicry. Ecology Letters, 13, 1494–1502. https​://doi. reveals ecological factors involved in the evolution of chemical org/10.1111/j.1461-0248.2010.01539.x defenses in mimetic butterflies. Ecol Evol. 2020;10:2677–2694. RStudio Team (2019b). RStudio: Integrated Development for R. RStudio, https​://doi.org/10.1002/ece3.6044 Inc., Boston, MA. http://www.rstud​io.com/ SCULFORT et al. | 2693

APPENDIX 1 Detailed list of sampled butterfly subspecies (n = 375 individuals), with number of females (n = 119) and males (n = 256) as well as provenance country (Brazil, Ecuador, Panama, or Peru)

Genre Species Subspecies Female Male Total Country Specialization

Agraulis vanillae luciana 1 3 4 Peru Generalist Agraulis vanillae vanillae 1 1 2 Panama Generalist Dione juno huascuma 1 2 3 Panama Generalist Dione juno miraculosa 5 8 13 Peru Generalist Dryadula phaetusa NA 2 6 8 Peru/Ecuador Generalist Dryas iulia moderata 14 24 38 Peru/Panama/ Generalist Brazil Eueides isabella dissoluta 8 11 19 Peru Generalist Eueides isabella eva 0 3 3 Panama Generalist Eueides isabella hippolinus 0 2 2 Peru Generalist Eueides lampeto acacetes 1 1 2 Peru Generalist Eueides aliphera aliphera 1 0 1 Brazil Generalist Eueides lybia lybia 0 4 4 Brazil Generalist Eueides tales calathus 0 1 1 Ecuador Generalist Heliconius aoede cupidineus 9 13 22 Peru Specialist Heliconius burneyi jamesi 0 1 1 Peru Specialist Heliconius charithonia vazquezae 0 2 2 Panama Generalist Heliconius congener congener 0 3 3 Ecuador Specialist Heliconius demeter joroni 2 0 2 Peru Specialist Heliconius doris doris 3 5 8 Peru Specialist Heliconius doris viridis 2 2 4 Panama Specialist Heliconius eleuchia primularis 0 2 2 Ecuador Specialist Heliconius erato cyrbia 0 1 1 Ecuador Generalist Heliconius erato demophoon 2 1 3 Panama Generalist Heliconius erato emma 1 4 5 Peru Generalist Heliconius erato favorinus 11 20 31 Peru Generalist Heliconius eratosignis ucayalensis 0 3 3 Peru Specialist Heliconius ethilla aerotome 5 16 21 Peru Specialist Heliconius hecale felix 0 2 2 Peru Generalist Heliconius hecale melicerta 2 4 6 Panama Generalist Heliconius hecale zuleika 0 1 1 Panama Generalist Heliconius hewitsoni NA 0 3 3 Panama Specialist Heliconius himera NA 2 3 5 Ecuador Specialist Heliconius melpomene aglaope 1 0 1 Peru Specialist Heliconius melpomene amaryllis 5 16 21 Peru Specialist Heliconius melpomene amaryllis*aglaope 1 2 3 Peru Specialist Heliconius melpomene rosina 1 3 4 Panama Specialist Heliconius numata arcuella 2 0 2 Peru Generalist Heliconius numata bicoloratus 4 15 19 Peru Generalist Heliconius numata lyrcaeus 1 0 1 Peru Generalist Heliconius numata tarapotensis 2 10 12 Peru Generalist Heliconius numata zobryssi 0 1 1 Brazil Generalist Heliconius pachinus NA 2 2 4 Panama Generalist Heliconius pardalinus butleri 1 1 2 Peru Generalist 2694 | SCULFORT et al.

Genre Species Subspecies Female Male Total Country Specialization

Heliconius pardalinus sergestus 3 11 14 Peru Generalist Heliconius sara magdalena 2 3 5 Panama Specialist Heliconius sara sara 16 22 38 Peru/Ecuador/ Specialist Brazil Heliconius telesiphe sotericus 0 3 3 Ecuador Specialist Heliconius timareta thelxinoe 0 1 1 Peru Specialist Heliconius timareta timareta 0 2 2 Ecuador Specialist Heliconius wallacei flavescens 2 8 10 Peru/Brazil Specialist Heliconius xanthocles melior 0 1 1 Peru Specialist Heliconius xanthocles zamora 2 0 2 Ecuador Specialist Philaethria diatonica NA 0 2 2 Peru Generalist Philaethria dido dido 0 1 1 Peru Generalist Philaethria dido panamensis 1 0 1 Panama Generalist Some species do not have subspecies name so it was “NA” assigned. Right column “Specialization” indicates whether subspecies are generalists (feed on wide panel of Passiflora plants) or specialists (feed on a restricted range of Passiflora plants).